More on Bayesian approaches to detection and attribution

Timely given events all over the place, this new paper by Mann et al has just appeared.  It's a well-aimed jab at the detection and attribution industry which could perhaps be held substantially responsible for the sterile “debate” over the extent to which AGW has influenced extreme events (and/or will do so in the future). I've argued against D&A several times in the past (such as here, here, here and here) and don't intend to rehash the same arguments over and over again. Suffice to say that it doesn't usefully address the questions that matter, and cannot do so by design.

Mann et al argue that the standard frequentist approach to D&A is inappropriate both from a simple example which shows it to generate poor results, and from the ethical argument that “do no harm” is a better starting point than “assume harmless”. The precautionary versus proactionary principles can be argued indefinitely, and neither really works when reduced ad absurdum, so I'm not really convinced that the latter is a strong argument. A clearer demonstration could perhaps have been provided by a rational cost-benefit analysis in which costs of action versus inaction (and the payoffs) could have been explicitly calculated. This would have still supported their argument of course, as the frequentist approach is not a rational basis for decisions. I suppose that's where I tend to part company with the philosophers (check the co-author list) in preferring a more quantitative approach. I'm not saying they are wrong, it's perhaps a matter of taste.

[I find to my surprise I have not written about the precautionary vs proactionary principle before]

Other points that could have been made (and had I been a reviewer, I'd probably have encouraged the authors to include them) are that when data are limited and the statistical power of the analysis is weak, it is not only inevitable that any frequentist-based estimate that achieves statistical significance will be a large overestimate of the true magnitude of the effect, but there's even a substantial chance it will have the wrong sign! A Bayesian prior solves (or at least greatly ameliorates) these problems. Another benefit of the Bayesian approach is the ability to integrate different sources of information. My favourite example of the weakness of traditional D&A here is the way that we can (at least this was the case a few years ago) barely “attribute” any warming of the world's oceans under this methodology. The reason for this is that the internal variability of the oceans is large (and uncertain) enough that we cannot be entirely confident that an unforced ocean would not have warmed up by itself. On the other hand, it is absurd to believe the null hypothesis that we haven't warmed it, as it has been in contact with the atmosphere that we have certainly warmed, and the energy imbalance due to GHGs is significant, and we've even observed a warming very closely in line with what our models predict should have happened. But D&A can't assimilate this information. In the context of Mann et al, we might consider information about warming sea surface temperatures as relevant to diagnosing and predicting hurricanes, for example, rather than relying entirely on storm counts.

Well-tossed word salad with French dressing

Bjorn put on a little conference last week in honour of our visit (or if you prefer alternative facts, we made sure our dates included the meeting). Mostly, it consisted of an up-to-date compendium of what people are currently doing in climate model development which was very interesting to us now we aren't based in a lab. Of course this work tends to be fairly incremental stuff so having a bit of a hiatus doesn't matter too much for those of us not actually working at the cutting edge of this stuff!

Interspersed between the science sessions were occasional “Cross-cutting invited presentations on the history, philosophy and sociology of Earth system science”. A fair chunk of this served only to reinforce my jaundiced view of the social sciences, but some was genuinely interesting and thought-provoking. In that respect, Wendy Parker presented an interesting discussion of the role of models and how we use them, arguing that rather than considering a model as a hypothesis to test (and which is inevitably false when examined in detail, as all models contain simplifications and approximations) we should instead form hypotheses about the adequacy of the model for a specific task, and aim to test these. I don't think this is revolutionary - surely many modellers already do think in these terms to some extent - but seeing it laid out explicitly in some detail was useful, I think.

There was some stuff about “consensus” and the role of the IPCC in contributing to the public understanding of science. I don't think this discussion really achieved a great deal. Thomas Stocker (who was present) gave a robust defence of the IPCC process but whatever your viewpoint, this is all largely irrelevant to the process of scientific research that most attendees focus on in their day jobs. I do think the IPCC is a bit of a dead weight sometimes, as an outsider it sometimes appears  that the authors consider their main role when responding to comments is to defend their first (public) draft against all criticism, and of course there's no independent editorial control over this (as there would be in most peer-reviewed publication). But on the other hand, even this supertanker can be observed to have slowly changed its direction over a period of time when we look back over a decade or so. Persistence, when combined with being right, generally wins out in the end. Thomas Stocker also presented some results from this analysis of the IPCC text from a readability point of view.  He was a bit apologetic about the fairly low scores achieved of 10-30 for the IPCC SPMs, though he did point out that not only was WG1 at the high end of this, but also the summary headline statements were generally significant easier reading than the texts. For calibration, scientific papers are around 40, quality papers 40 and tabloids 50 on the Flesch Reading Ease score that was used. 

The slide that motivated the blog title was a dense screed of text from one of the social scientists which, when I analysed the final sentence with the Flesch Reading Ease formula, achieved the notable score of -21.5. I think this probably means that it can be understood by no-one, even the author. However she did partially redeem herself by referring to the silly 1.5C limit stuff as fake science, relying as it does on fantasy technology that doesn't yet exist (at any meaningful scale). I'm still rather disappointed by the alacrity with which the IPCC has jumped at the opportunity to write a report on this, despite the utter futility of the exercise.

I see I haven't really commented on the science. Um, science was being done, by lots of people, in a number of different directions. A decade, in the context of “decadal prediction”, still means 2-5 years, this being about the limit of any sort of measurable (let alone useful) prediction skill. My decadal prediction is that they'll stop calling it decadal prediction in about another decade :-) There is lots of carbon cycle modelling which I have never really got that excited by. I know it's important for determining how the climate will change (as a function of emissions) but it still seems a bit ad-hoc and empirical to me. Paleoclimate research is increasingly valuable for testing models, though I'm not sure how it will all fit into the new IPCC chapter structure that is on the verge of being approved. But as above, that's not about the science per se but merely about how it's summarised. Should I say something more? Well, if anyone has a specific interest piqued by the program, ask away. I think presentations may also appear on the website at some time. By the way, having made a rather late decision to come, Jules and I were merely attendees with no presentations of our own.

Practice and philosophy of climate model tuning across six US modeling centers

Paper with the above title just appeared in GMD. Despite being a European English-language journal we welcome Americans and even Americanisms, so I'll quote the title as written rather than as it should be :-) In this paper, Gavin has nicely summarised (or perhaps I may say, summarized) how approaches to model tuning vary throughout the US climate science community.

It's a slightly unusual manuscript type for GMD in that it doesn't present any technical advances (such as parameter estimation techniques, examples of which have been published in GMD previously) but instead describes the rather more ad-hoc hand tuning that model developers currently do. As such it generated some behind-the-scenes discussion as to how best to handle the manuscript within the GMD framework. We at GMD have always seen our role as enabling rather than constraining the publication of modelling science, and were already considering the concept of “review” paper types which survey a field rather than notably advancing it, so this was an opportunity rather than problem for us. The reviewers also made constructive comments which my job as editor fairly straightforward.

A major point of interest in the paper (and in model tuning generally) is to what extent the models have or have not been tuned to represent the 20th century warming. This has significant implications for how we would interpret their performance and potentially use the observational data to preferentially distinguish between models. Gavin has always been quite insistent that he doesn't use these data:

I got tired of explaining that climate models don't tune to 20th Century temperature trends. So we wrote a paper.https://t.co/EgDY8Cvogm pic.twitter.com/YR4MkZjqZs

— Gavin Schmidt (@ClimateOfGavin) September 1, 2017

and I certainly have no reason to doubt his claim. On the other hand, this Isaac Held post on tuning is also worth reading. In that post, Isaac Held argues that the warming is probably baked-in to some extent in the way that the models are built and evaluated during their construction. On balance I think I prefer Isaac's way of putting it to Gavin's, but it's a nuanced point. Certainly there is no question that modellers do not repeatedly re-run 20C simulations, tweaking parameter values each time until they get a good fit to the observed record. So if this is what people are envisaging when they discuss the topic of “model tuning” then Gavin is certainly correct, this simply doesn't happen. And I'm happy to believe that some modelling teams don't run the 20C simulation at all until the very end of the model development phase, and simply send their very first set of simulation results to the CMIP database. But I've seen for myself that some groups have sometimes done these simulations at an earlier stage, and on seeing a poor result, have gone back and redesigned some aspects of the model to fix the problems that have arisen (these are likely to be more specific than just “the wrong trend”). And even beyond this, it's a bit of an open question to what extent the tuning that is done to individual model components is truly independent of our knowledge about the recent warming which constrains our estimates of various aspects of model behaviour. But given the limited nature of any such tuning, (and indeed the limited agreement between models and data!) perhaps it's a close enough approximation to the truth to just call them untuned.

Hamburg revisited

Yes we're in Hamburg again, which at least partially explains the ferry. We are back at the Max-Plank Institut für Meteorologie for another visit courtesy of Bjorn Stevens and Thorsten Mauritsen. Who is not Mauritian despite the attempts of my spellchecker to make him so! Going by ferry enabled us to bring bicycles (in the new stealth camper van) 

and rent an apartment a little way out of the centre without bothering too much with public transport (which is of course actually very good, but never as good as a bike), and also tied in with a brief visit to jules' brother and family who are conveniently located along the way in the Netherlands.

(Token photo of Dutch family cycling)

Wrong side driving in a right hand drive van with limited visibility wasn't anywhere near as fraught as I'd feared it might be and we got to Hamburg in one piece thanks in no small part to Google's navigation on our phones. Damn the evil interfering EU and their abolition of roaming charges! Pavement cycling on all the crazy lanes in Hamburg is considerably more challenging, I have adopted the habit of tucking in behind someone who seems to know what they are doing and just following their lead. Fortunately the hamburgers seem much more tolerant of each other than the british would be in similar circumstances.

So we are back in science mode and more blogging on random topics should be forthcoming. Hooray, I hear you shout. Well actually I didn't. But I'll do it anyway.

[jules' pics] ships - Harwich to Hook of Holland

So much more civilised than flying... especially when transporting a tandem.

--
Posted By Blogger to jules' pics at 8/26/2017 08:43:00 PM

Bishop's Castle Tandem Triathlon

We were aware of this annual event many years ago before we went to Japan. However, back then I didn't run and jules was not that regular a swimmer, so we never got round to entering it. The last couple of years since returning from Japan it has clashed with our local club triathlon so it was only this year that we finally got round to doing it for the first time. It's a very informal event and the format is slightly non-standard in that the pool swim (1km), tandem ride (32km), run (10kmish) is followed by a final 5km tandem ride back to the event centre as the (trail) run is situated in a hilly forest some way out of town.

We drove down to Bishop's Castle the day before, not having worked out that this meant “enjoying” Friday evening rush hour on the M6...though it didn't turn out too bad as it happened. My Airbnb arrangement was a comedy of errors due to various communication problems and minor misunderstandings but it all turned out very well in the end and we found a decent pub dinner for pre-event calorie-loading - the traditional high-fat high-beer version as pioneered by yours truly for previous marathons (e.g. 1, 2). The airbnb had a little kitchen so we made quite a success of breakfast too. Theresa May take note!

Our start time was 10:52 which gave plenty of time to digest the bacon, black pudding, potato scone, egg, tomato, beans, mushroom, cereal, croissant, banana, yoghurt (albeit not all on the same plate). Oh, and we also managed a good look at the chaos taking place in the pool before jules had to jump in and join it. There were 4 lanes each with about 3 swimmers splashing up and down who had been theoretically laned according to their predicted swim times, but it seems they were mostly lying or, perhaps we may charitably say, mistaken as to their prowess. Jules ended up overtaking her breaststroking lanees multiple times but if anything this seemed to spur her on to a sub-19 min time, comfortably ahead of expectations. Then she made a quick switch of googles for specs and on the with the shoes and helmet and off we were on the bike.

The bike leg was great, it was fun to go properly fast again on a fairly flat course on A and B roads. There were plenty of tandems ahead of us to chase down around the course and one quite steep hill that we had to do twice (the route being a lap and a bit). Up the hill for the second time and I had to jump off and head into the forest for a moderately hilly run. That was really hard under what was by then a very hot mid-day sun. Even trees don't provide much shade when the sun is vertically overhead! There were times that I seriously considered stopping/walking/lying down in a ditch but fortunately water bottles handed out on course stopped me from expiring. I didn't even get a rest while changing shoes as due to the double transition I'd decided to just ride in running shoes and old-fashioned toe clips (which worked out fine, it didn't really handicap us on the bike). Back on the bike and downhill to the finish was a bit of a blur and then we forced down some food and drink before watching the later teams come in.

 

Perhaps "lying down in the sun while people ride past" would be more precise!

(top two pics from Sandy Plenty, bottom one by jules)

Chip timing and a live-updated web page (apparently for the first time this year) meant we could immediately see how well we had done and who of the remaining competitors was likely to be close...initially we were the first mixed team and 2nd overall....and then one more male team came in ahead...but all the mixed teams were a bit slower! Making us the national mixed tandem triathlon champions for 2017. Possibly world mixed tandem triathlon champions too, as this may be the only event of its type in the world. And we were first vet (over 40) team of any gender too. Jules has been basking in her multiple QOMs on strava, even though I somehow still managed to complete the first lap some 8 seconds faster than she did :-)

To be fair, all the fastest women are on tandems, this is clearly a route that only tandemmers race round (it's not an official RTTC course). So we don't feel too guilty about the ranking. We're hoping to return to defend our title...

The Verdict

I'm probably supposed to be doing some climate science, but that's not been top priority recently (might be different if someone was paying me to do it). Actually there is some slow progress and there will probably be something to report eventually. But it's all gone a bit boring in advance of the next CMIP experiments and IPCC report.

So, this is more politics. After the arbitration, the verdict. Not mine, but the country's. Don't shoot me, I'm just the messenger. I greatly enjoyed Thursday night, starting with the exit poll that no-one really believed until the results came in more or less confirming it. The local constituency went as expected, though at least the Green candidate saved his deposit with one of the best results they obtained around the country. The adjacent LibDem fared little better. The surprises were elsewhere.

To be fair, May did warn us that if she didn't get a majority we'd face a coalition of chaos propped up with terrorist sympathisers. The only point she forgot to mention was that she'd be leading it. It seems pretty much an ideal situation, a humiliated and ineffectual Tory govt limping on for a few months until the inevitable failure of the brexit process finally puts them out of their misery. Their eagerness to jump into bed with stone age bigots really does tell us (anyone who was still wondering) what sort of people they are, and is especially poignant after having been dealt such a spanking by an increased progressive vote from the 18-24 age group. That's one way to not learn a lesson.

Most people on the mainland are pretty uninterested in NI politics, myself included, but when they start to realise quite how unpleasant the DUP are, there may be significant opposition to any alliance. It's not just a matter of being effectively the political wing of the UDA and UVF. They are also reactionary creationist homophobic climate change deniers. In fact they are so toxic that the Scottish Tories have already threatened to break away if the DUP are allowed to influence Govt policy. And if they are not, then what exactly is the nature of their alliance? It will hardly reduce tensions in NI either, where the power-sharing structures are currently struggling to survive due to an ongoing scandal that the DUP are up to their necks in. If the UK govt so blatantly allies itself to one side and re-imposes direct rule there's likely to be unrest to put it mildly. May is of course completely tone deaf to any and all concerns, as her “carry on regardless” approach shows, though she won't be able to escape from reality indefinitely. We've already seen that she can't even reshuffle her ministers.

I'm not sure that a Labour govt would have been preferable really, since then they'd have had to carry the can for brexit and the Tories would then have been able to argue that they would have made more of a success of things. Of course they cannot, the unspoken truth which has poisoned the whole political process over the last couple of years is that brexit cannot possibly succeed. It is the Kobayashi Maru of modern politics: a test that has no positive outcome, the only way to win is not to play:

“The objective of the test is not for the cadet to outfight or outplan the opponent but rather to force the cadet into a no-win situation and simply observe how he or she reacts.”

The Leave campaign was allowed to get away with saying different (and contradictory) things to different people but as soon as anyone tries to put together a coherent strategy, it is obvious that there's no good outcome. The DUP has a particularly brilliant strategy of simultaneously demanding both no special deal for NI and no hard border with the Republic. Easy to say in a manifesto, not so easy to work out what it actually means in concrete terms. So far, no-one has exactly covered themselves in glory over brexit plans - perhaps the LibDems and Greens are the closest to having a rational response - but we must remember it was the Tories, and those who voted for brexit, who created this situation. They broke it, they own it.

The desperate scramble to put together this Maydup Coalition in time for the brexit negotiations is the icing on the cake: the timing of the election is entirely the responsibility of the Tories who forced through the invocation of Article 50 immediately before calling it. They can't even replace May, there's no way they can afford the time for a leadership election now and any new leader would face the inevitable (and accurate) criticism that they had no mandate for their views. With their election campaign having been so focussed on May as the “strong and stable” leader, they seem to be stuck with her for the time being at least. Though it's obvious enough that the plots are well underway, it is only a matter of choosing the first available moment to plunge the knife.

In another installation of “what has the EU done for us” (and returning briefly to science again), they are pushing hard for all research outputs to be open access by 2020. Hooray, we (supposedly) won't be EU members by then so can hide our research behind paywalls. 

Arbitration

My reader is probably waiting with bated breath for my views on the forthcoming election. In the blue corner, Teresa “strong and stable” May (or should that be “u-turn when you want to”). In the red, Jeremy “don't mention the brother” Corbyn. In the ignored corner, Nutter and Failing and a few others.

Of course, it's all about brexit, so there hasn't been any sort of meaningful debate about this. Both tories and labour are rushing headlong for the most catastrophic outcome they can possibly engineer, and there isn't a fag-paper of difference between them on anything substantive. Corbyn promises better employment protection and May less red tape but these are not really issues of how and why we leave the EU, rather what we do afterwards. The Labour vision may be marginally more attractive but that's basically a question of what colour deck-chairs you prefer on the “Titanic Success”.

It's important to realise, there is no such thing as  a “good brexit”. The only reasonable brexit would be something functionally indistinguishable from the status quo, which both sides have ruled out. The choice is between a bad brexit, a worse brexit and a catastrophic brexit, with all the smart money on the latter. All competent experts have repeatedly pointed out the huge problems that brexit will bring, including but not limited to our European flights (there's no agreement for anything post 2019 and timetables will have to be designed well in advance of that), the operation of our nuclear industry (including such details as medical isotopes), the huge customs problem at Dover/Calais for which the infrastructure does not exist and simply cannot be built in time, the Northern Irish border which will likely spark off unification violence, the harm to our financial industry, the fact that we aren't even normal WTO members in our own right and negotiating that will take agreement from the other 162, the 759 separate agreements with 168 countries that need to be renegotiated in the remaining 661 days etc. The whole thing is idiotic nonsense and the failure of most of our politicians to say as much in plain terms is a gross dereliction of their duty.

In my opinion, the most likely outcome by some way remains a year or so of increasingly acrimonious negotiations or rather arguments, followed by a collapse of the process and long period of recrimination. This national humiliation will come at great cost of course, not just economically but also politically, culturally and socially, as we are already starting to see. Lots of people are starting to bleat about the entirely predictable consequences. I'm intensely relaxed about the poor farmers, since just about every field round here had a “Vote Leave” placard this time last year. They of all people should realise that they will reap what they have sown!

And all for what? Even though it was all about “taking back control”, no-one is prepared to make any promises about immigration anyway. For while the EU was always the convenient excuse for the large-scale immigration that govts of all stripes have encouraged over recent years, it was never actually anything more than that. They could have reduced immigration substantially had they wanted to, but they saw the obvious economic benefits of it and rather than arguing honestly in favour, passed the buck on to the EU. That is no longer possible, so we get vague ambitions which are quite clearly meant as nothing more than a dog-whistle to UKIP voters who are presumably thought to be too stupid to realise that there isn't a plan to actually do anything.

I've been enjoying the way the Tory campaign has been falling apart. “Strong and stable” has been laughed out of existence, it doesn't seem such a clever idea now to base the campaign on May's personality when it is obviously so brittle and unpleasant. Amusingly, the local election leaflet which eventually plopped through the letter box just a couple of days ago features May more centrally than the candidate himself, which should lose him some votes but I'm sure won't affect the result. As for the central conceit of May being the only person capable of negotiating brexit, if anyone seriously thinks she'll actually be PM on the day we leave the EU, I feel a wager coming on.

The only mainstream party with anything approaching a sane policy on brexit is the LibDems, so I'll be supporting them. Or rather, I would be, but they have agreed a local pact with the Greens in a neighbouring constituency (Harrogate and Knaresborough) that they won't stand here and the greens won't stand there. Think they've got the better of that bargain as H&K is at least a possible LibDem seat though not one of their top targets. So I'll be voting Green instead, the candidate seems to have made a very good impression locally (though we missed the local hustings, being in Hamburg at the time). The Greens have a very similar policy to the LibDems anyway and it wouldn't matter who stood here, the local “pig in a blue rosette” is guaranteed to be re-elected even though he was a staunch remainer this time last year.

[jules' pics] Pond

Hornby Castle belongs to someone who I was at primary school with for a few years before he went to Eeeeton. I probably only remember this because his grandmother joked about having to take out a second mortgage on the castle in order to afford the uniform. Oliver won't remember me, of course, especially as he is a few years younger, but he sometimes opens his castle gardens up to visitors. Luckily, there was time to walk round most of the garden before the rain started, and just as it got going, his aunt gave a talk about the history of the castle, which was quite an achievement as she started in the 11th century and used no notes.

--
Posted By Blogger to jules' pics at 5/21/2017 05:45:00 PM

Peak performance

A few weeks ago, I took part in the annual Three Peaks Race (spring version). In contrast to the autumn version that jules and I helped to marshall, participants in the spring event do not have to carry a bicycle but may walk or jog unencumbered bar a waterproof layer and bar of chocolate. Not too much walking though, as the time cut-offs are fairly severe. A 4 week gap from Manchester Marathon was plenty of recovery time, at least that was the theory. As the date approached I realised I wasn't really that motivated to aim for an optimal performance, and just set myself the target of getting round without too much of a struggle.

The race, as you might have guessed, takes in the famous three peaks in the Yorkshire Dales, which are conveniently located just a few miles up the road from us. So I had no excuse for not knowing what I had let myself in for, and had done a couple of "2 peaks" training runs, though never yet covered the full race distance. Recent weather had been dry making for very fast and easy conditions, with few of the usual bogs. But it's still 38km and 1600m of climb and descent. 

There were several other club members entered, including one who was probably going to be a bit quicker than me. I set off steadily and as expected he gradually pulled away over the first hill, eventually disappearing out of sight. Very much to my surprise, I caught him up around half way, he had a bad day and finished rather slowly.

Jules came to support and hand me my water bottle at the pit-stops where the course crosses roads, for which I was grateful on an unusually hot day for April.  She also took a couple of photos of course:

Runners heading towards Ingleborough

What do you mean "that's not much of a peak"?

I had a reasonably successful day, finishing comfortably under 4 hours though really I think most people would hope to take less than an hour more than their marathon time, especially with such good conditions. The women's record was well broken by Victoria Wilkinson, the men's record was not after the leader managed to head off the wrong way down Pen-y-Ghent. I think one or two veteran records went too.

In honour of the event, I have composed a haiku:

Ah Pen-y-Ghent ah
Ah ah ah Whernside ah ah
Ah Ingleboraaargh...

With apologies to Basho, although interestingly it seem that one of his most famous poems was not actually written by him.